Multi-Threaded Super-Pi Contest

8

87dtna

Active Member
Well looks like I max out at 205 BCLK on this $115 asrock board LOL. All voltages maxxed out pretty much but I got 4.3ghz stable enough to run super PI 32m( I know it says 4.1 but I changed multi to 21 with OC tuner)

superPI43ghz.jpg


Was hoping I could get 4.4ghz out of it to try and beat Jet's 10.4 seconds, but oh well I guess. Vcore for 4.3ghz took a staggering 1.5875 to get stable! My TRUE cpu cooler still keeping it cool as a cucumber around 70c core temps and under 50c actual CPU temp. Max spec is 97c. For that kind of voltage I'm impressed, TRUE ftw!
 
Last edited:
Jet

Jet

VIP Member
87dtna said:
Well looks like I max out at 205 BCLK on this $115 asrock board LOL. All voltages maxxed out pretty much but I got 4.3ghz stable enough to run super PI 32m( I know it says 4.1 but I changed multi to 21 with OC tuner)

superPI43ghz.jpg


Was hoping I could get 4.4ghz out of it to try and beat Jet's 10.4 seconds, but oh well I guess. Vcore for 4.3ghz took a staggering 1.5875 to get stable! My TRUE cpu cooler still keeping it cool as a cucumber around 70c core temps and under 50c actual CPU temp. Max spec is 97c. For that kind of voltage I'm impressed, TRUE ftw!
Click to expand...

Did you raise the other voltages? (I have a sub 10 second run, but using ASRock's tool made it not like it, so I'm trying to find a solution for that ;) )
 
8

87dtna

Active Member
Yeah, VTT seemed to help the most but it seemed to max out at 1.49v. If I set it to 1.50 I get freeze ups. I took PLL to 1.90, that seemed to help slightly too but more didn't make any difference so I took it back down to 1.90. PCH didn't help anything. It's only a $115 board, I was expecting to have to fight with it to get 200 BCLK stable but it did that on all stock settings! Well except Vcore. My cpu does stock voltage up to 3.6ghz, 1.35 at 3.8ghz, 1.40 at 4.0ghz, and 1.48 at 4.2ghz.

I'm guessing 4.4ghz would need at minimum water cooled if not No2 or dice. 4.2ghz is stable at 1.49 Vcore, but like I said it took 1.58 to get 4.3ghz stable. Not going any higher than that because that was crazy already LOL.



Oh, and can you update my 32m time? Thanks
 
Last edited:
2048Megabytes

2048Megabytes

Active Member
Ha! Ha! Ha! :D I tried to overclock my Athlon 7750+ and I think I fried it. After I attempted to overclock all I could get into was my BIOS and I ran Memtest86+ with the Athlon 7750+. It worked for 7.5 hours with Memtest86+, no errors, but when I would try to start into any operating system I would just get black screen system hangs.

So I changed out my processor and now everything seems to be fine. Guess I am with my good old Athlon 4600+ Dual-Core again. I think I will do a Y-Cruncher test on the Athlon 4600+ processor and see what I get.
 
Last edited:
8

87dtna

Active Member
2048Megabytes said:
Ha! Ha! Ha! :D I tried to overclock my Athlon 7750+ and I think I fried it. After I attempted to overclock all I could get into was my BIOS and I ran Memtest86+ with the Athlon 7750+. It worked for 7.5 hours with Memtest86+, no errors, but when I would try to start into any operating system I would just get black screen system hangs.

So I changed out my processor and now everything seems to be fine. Guess I am with my good old Athlon 4600+ Dual-Core again. I think I will do a Y-Cruncher test on the Athlon 4600+ processor and see what I get.
Click to expand...

What??? What settings did you try/change?!!!

Did you try resetting the bios first before the CPU change? If you change the CPU the bios normally resets itself so thats why it worked again. Sounds like the settings you tried were simply not stable. Seriously doubt you fried the CPU, especially if you did not change any voltage settings.
 
2048Megabytes

2048Megabytes

Active Member
I reset the BIOS to the fail safe settings and it still wouldn't work correctly. The Athlon 7750+ Dual-Core will go back in to see if it works. I can't exactly remember what BIOS settings I tinkered with but I didn't bump the voltage up by very much. I was mainly messing with the multiplier on the processor and the north bridge.
 
2048Megabytes

2048Megabytes

Active Member
I put the Athlon 7750+ back into my motherboard and got the same results. When I try to start into any operating system I just get black screen system hangs.

Switched out my processor to the Athlon 4600+ and BAM, everything seems to work fine. Guess I'm back to the old Athlon 4600+. Bummer, since I only owned the Athlon 7750+ for about two days. Oh well, better to laugh about it. :D Life is too short to worry about one computer chip.
 
8

87dtna

Active Member
Wow umm that doesn't really make any sense. What did you do to the north bridge? Just change the multi or did you change the voltage?

What Vcore did you set it too?

If I had an AM2+ motherboard I would ask you to send it to me to test but I don't.
 
Jet

Jet

VIP Member
32M @ 4.3

fastest32.png


256M @ 4.2

fastest256.png


512M @ 4.2

best512M.png



Poke, I have some WRs for the non-SuperPi calculations of you want those--I'll post them later. That 512M time is a fastest time as well, looking at the charts.
 
Last edited:
poke349

poke349

New Member
Jet said:
Poke, I have some WRs for the non-SuperPi calculations of you want those--I'll post them later. That 512M time is a fastest time as well, looking at the charts.
Click to expand...


Yeah, let's see them. Be glad to update them. :D



So I just finished running a whole set of benchmarks on my workstation using the latest version (with the efficiency %s)...

The larger the computation, the better the efficiency, so I was gonna see what 8 cores on 64GB of ram can do...
And it couldn't break 96%... (My i7 does it easily... even at much smaller sizes...)

Any ideas?

Same OS, same service pack, no background programs... 8 virtual cores each...
It should have nothing to do with memory bandwidth because the efficiency measures how much cpu time is spent (as visible to the OS) - which includes all the delays for waiting on ram...


Digits - Efficiency

150,000,000 - 88.3461 %
200,000,000 - 89.6781 %
250,000,000 - 90.3781 %
400,000,000 - 92.1596 %
500,000,000 - 92.8779 %
750,000,000 - 93.6555 %
1,000,000,000 - 93.9021 %
1,200,000,000 - 94.2294 %
1,500,000,000 - 94.6377 %
2,000,000,000 - 94.8800 %
2,500,000,000 - 94.9631 %
4,000,000,000 - 95.2968 %
5,000,000,000 - 95.2680 %
7,500,000,000 - 95.6914 %
10,000,000,000 - 95.9038 %
12,000,000,000 - 95.4064 %
 
Jet

Jet

VIP Member
just upgraded from a single pass 120mm radiator to a double pass 240mm--so I'll hopefully be getting some benching in when I get back from a trip in mid-January!
 
poke349

poke349

New Member
Just a little bump and a heads up on something...

I need some opinions on what I should do:

The next version (v0.5.x), is faster than the current release. (It's nowhere near ready yet, so don't get excited...)
I won't say how much faster, but it's enough to wreck havoc on competition threads like this if people started posting benchmarks using it.

What should I do?

  1. Delay the release of v0.5.x for as long as possible.
    Pros: Gives v0.4.4 a longer lifetime. Avoids wrecking havoc on competition threads for the meantime.
    Cons: Other improvements to the program won't get out. (The new validation scheme being the biggest of them all for now.)

  2. Branch the program into two versions: "y-cruncher HPC" and "y-cruncher Benchmark". (Not necessarily those names, but something along those lines.)
    Then require that "y-cruncher Benchmark" be used for competition threads.
    Pros: Allows new improvements to be done without wrecking havoc on existing benchmarks.

    Cons: I'm 90% sure this will confuse the hell out of people... This is also a little too much work for me since I'll have to maintain double the code.

  3. Release it anyway and hope everything turns out okay.

  4. Notify all threads like this and let them decide what to do.

Basically, there's a conflict of interest right now. "Make a benchmarking program." vs. "Make an HPC program."

Since "benchmarking" requires consistancy...
And "HPC" implies making it faster and better...


As of right now, I'm leaning towards option 1 since it gives me the most flexibility to do things.

And if we decide to go the consistency route, how will new tech like AVX be incorporated?
 
S

Sam_VDC

New Member
87dtna said:
Ah man now I gotta re-run my 256 and 512 at 4.3ghz LOL.

What Vcore are you running to get 4.2ghz stable?
Click to expand...

Base on that setting, 1.39575 for manually.....and actually Auto Vcore rock solid as well.
With Auto Vcore, load P95 sFFT from CPU-Z is 1.376v
 
Last edited:
Drenlin

Drenlin

Active Member
So it's just sinking in that I got my arse handed to me by a netbook. I thought it was funny at first, but now....I think I need to get my laptop running again!
 
Last edited:
2048Megabytes

2048Megabytes

Active Member
Drenlin said:
So it's just sinking in that I got my arse handed to me by a netbook. I thought it was funny at first, but now....I think I need to get my laptop running again!
Click to expand...

The Intel Atom processors don't have very much processing power. The old Celeron processors are very slow. If you have a Windows installation disk you could build yourself a decent desktop system with an Athlon II Dual-Core Processor for around $350. That would be a huge upgrade from your present technology.
 
You must log in or register to reply here.
Top